A Rs 6,000 crore index fund must buy a new index entrant at a 2.4% weight on the rebalance date, Rs 144 crore of a stock that trades Rs 90 crore a day, and traders have bought ahead. What does front-running cost investors, and could the fund do anything differently?

Buying at the close costs investors about Rs 5.76 crore, 4% of the purchase or about 10 basis points of the fund, roughly half a year's expense ratio, and it never shows as tracking error because the index pays the same price. Only the part of the run-up that reverses is a cost. The fund can buy earlier or later more cheaply, but each route is a tracking bet. The realistic fixes are a small pre-agreed tracking budget and pressure on index design.

If everyone knows a large caterer must buy all the mangoes in a market on a fixed date, traders will buy mangoes first and sell them to the caterer at a higher price. Index changes are public and the buying is compulsory, so traders buy the stock after the announcement and sell it to index funds at the rebalance close. Here the fund alone must buy 1.6 days of normal volume, and all trackers together about 12 days. This is legal anticipation of public information, not illegal front-running, which uses knowledge of a client's order before it is placed; keep the two apart in the interview.

Not all of the run-up is a cost. Inclusion brings permanent buyers and better liquidity, so some of the rise stays; only the part that reverses was paid away. Buying Rs 144 crore at 108 and valuing it at 103.68 ten days later is a loss of about Rs 5.76 crore, which is about 9.6 basis points of the fund. Against a 0.20% expense ratio, one inclusion costs investors about 48% of a year's fees, and none of it appears in tracking error because the index buys at the same price. The cost shows up instead as an index return slightly worse than the stocks' underlying value.

It could change when it buys, but every alternative is a bet. Spreading the purchase over the three days before, at about 106.5, would have overpaid only about Rs 3.81 crore and beaten the index by about Rs 2.03 crore. Waiting and buying over five days after, at about 104.5, would have overpaid only about Rs 1.13 crore. Those savings are hindsight: ex ante the fund is choosing to differ from its index, and if the stock had kept rising it would have lagged by the same mechanism. An index fund is hired to track, not to forecast.

Three things, in order of how much they help. Agree a small, written tracking error budget for index changes, so the dealer can pre-position part of the trade and take liquidity offered by stock-exiting sellers, accepting a measured difference from the index. Use the closing auction and crossing with other funds' sell orders where possible, which cuts market impact even if it cannot remove the run-up. And press the index provider, alongside other passive managers, for designs that blunt the game: longer and less predictable implementation windows, or staged entry over several rebalances. The last helps every tracker and is the only fix that removes the edge rather than sharing it.